What are the main ways to interact with Scarif?

Scarif offers two primary interfaces: an MCP server for seamless integration with compatible clients and a command-line interface (CLI) for both interactive querying and single-query execution.

How do I add documents to Scarif for searching?

Simply place your `.md` files into the designated `data/` directory within Scarif's structure. Scarif will automatically discover and index them the next time its search component is run.

Can I customize Scarif's search algorithm behavior?

Yes, Scarif allows you to fine-tune the BM25 search algorithm by configuring parameters such as `k1` (term-frequency saturation) and `b` (document-length normalization) directly in its `search.py` file.

Scarif is a lightweight BM25 keyword search engine designed for local Markdown files, providing fast and relevant search results through an MCP server or command-line interface.

Scarif

Name: Scarif
Author: Mandoa-Labs

byMandoa-Labs

0•

Developer Tools

Learning & Documentation

Provides a lightweight BM25 keyword search engine for local Markdown files, accessible via an MCP server or command-line interface.

Scarif is a versatile tool designed to bring powerful BM25 keyword search capabilities to your local Markdown document collections. Easily index your `.md` files by simply placing them into a designated `data/` folder, then query them using any MCP-compatible client like Claude Code, or directly through its intuitive command-line interface. Ideal for navigating personal knowledge bases, project documentation, or research notes, Scarif offers fast and efficient information retrieval with tunable BM25 parameters for optimized search results.

Key Features

01MCP server for seamless integration with compatible clients

02Configurable BM25 parameters (k1, b) for fine-tuning search behavior

03Interactive and single-query command-line interface

04Automatic indexing of Markdown files from a local directory

050 GitHub stars

06BM25 keyword search algorithm for relevant results

Use Cases

01Integrating local search capabilities into AI agents via MCP

02Querying local documentation for quick information retrieval

03Searching personal knowledge bases stored as Markdown files